PSI - Issue 13

L A Wray et al. / Procedia Structural Integrity 13 (2018) 1768–1773 L A Wray/ Structural Integrity Procedia 00 (2018) 000 – 000

1769

2

intersect at an angle (Mills & Eliasson 2006), (Zhou et al. 2012). Coating mechanical properties (Song et al. 2011), (Lee & Kim 2005) and thermo-mechanical properties (Zhang et al. 2005) have been measured and associated with cracking performance. Other investigations have considered strains caused by coating cure (Knudsen et al. 2006) and moisture absorption (Negele & Funke 1996) as possible driving factors for crack initiation. The coating application conditions for WBT coatings are not ideal and it is known that the coating dry film thickness (DFT) is difficult to control. This results in a final thickness that can vary significantly from the manufacturer’s specification. In their report Knudsen et al (2006) presented the intended and actual DFT of their test specimens and acknowledged that even under laboratory conditions there was a difference of up to 20% in measured DFT. The work of Zhang et al (2005) recognised that cracking resistance of epoxy coatings is dependent on DFT. In this study the effect of weld geometry and DFT on the thermal fatigue lives of coatings on welded steel was investigated. This was achieved firstly by measuring the mechanical and thermal properties of the coating including strength, toughness and thermal expansion. Coating durability was assessed by thermally cycling welded steel T sections with different weld geometries and different coating thickness. Crack development was monitored throughout the tests. Finite Element Modelling was employed to calculate thermally induced strain fields in coatings the weld. 2. Experimental Methods 2.1. Material Mechanical and Thermal Properties The WBT coating selected for study was a highly filled epoxy. Young’s modulus, strain to failure and fracture stress were measured on free film samples. These were produced by spraying the coating on to PTFE coated glass to produce a film of nominal DFT 300µm. This was cured at ambient for 7 days and post-cured at 100°C for 2 days to ensure the properties of the coating matched those achieved in service. Dogbone samples, Figure 1, were cut from the sheets after 2 days of ambient curing. Tensile testing was carried out at temperatures between -10°C and 60°C and stress-strain data extracted as described in an earlier paper (Wu et al. 2016). Young’s modulus was determined from each stress-strain curve using a method based on ISO 527-1.

Figure 1: Dimensions (mm) of free film samples

Thermal properties were measured also using free film samples. The glass transition temperature (Tg) was determined using Differential Scanning Calorimetry on a section of thin film. Coefficient of Thermal Expansion (CTE) was measured on free film cylindrical samples approximately 5mm in diameter and height. These samples were tested using a Thermo-Mechanical Analyser under the process outlined by Wu (2015). The Tg of the coating was found to be 69±2°C. The results in Table 1 show that coating modulus decreases with increasing temperature while the strain to fracture increases. The CTE of the coating is 5 times larger than that of the steel substrate.

Table 1: Mechanical and thermal properties of coating (measured) and steel substrate (Wu 2015), (Cverna F. 2002).

Poisson’s Ratio

Temperature (°C)

Modulus (GPa)

Strain to Fracture (%)

CTE (x10

-5 )

Material

Steel Substrate

All -10

207 6.3 5.2 2.5

45

0.30

1.2

0.27 0.34 0.95

Coating

23 60

0.31

6.0

2.2. Thermal cycling Sample Preparation Welded T-sections were produced to represent the geometry of stiffeners found within WBTs. The sections were manufactured by MIG welding 6mm thick steel plates to give the sample geometry in Figure 2A. To investigate the effect of weld fillet geometry on coating life one of the welds on each T-section was milled to a constant 2mm radius.